Why do poor whites do so badly at school? For example, only 32.3% of white British eligible for free school meals get five or more "good" GSCEs, whereas 52.% of Asians eligible for them do so and 48.2% of blacks do (Excel table here, via Sam).

Here's a theory.It rests on two premises:

- ability to pass exams and earn a living is highly correlated between parents and children, as Greg Clark suggests. For our purposes, it doesn't matter whether the transmission is via genes or culture.

- ethnic minorites are more likely than white British to be underpaid relative to their abilities. This might be because they suffer labour market discrimination (pdf), or because immigrants are more likely to be over-educated for their jobs than natives.

These two premises alone would generate a pattern in which poor white British do worse at school than ethnic minorities. This is because many poor minority children will come from good-ability parents who happen to be poor, whereas the white British will be more likely to come from lower-ability parents. And insofar as ability is inherited, minorities will then tend do better at school.

Now, neither left nor right will like this. The left will object to the assumption of a high inter-generational transmission of ability, and the right to my highlighting of racial discimination in labour markets. And managerialists will object to the apparent fatalism.

But it doesn't matter what you like. What matters is what's true. Is this theory?

Perhaps the strongest evidence against it pointed out by the Education Select Committee (par 44 of this pdf) - that there are massive regional variations in poor whites' attainment. For example, in Peterborough only 12.6% of whites eligible for free school meals get five good GSCEs, whereas in Lambeth almost half do. This huge discrepancy, says Adam Bienkov, suggests that where poor whites do badly it is because of poor schools rather than a lack of inherited ability.

Undoubtedly, this is part of the story. But I'm not sure it's all.

For one thing, in most (not all but most) of those local authorities where poor whites do well at school, poorer ethnic minorities do even better. For example, in Lambeth 49.6% of poor whites get five good GSCEs, but 57.7% of poor minorities do. (Annex 2 here). And even in Lambeth, poor whites do worse than the average poor Asians across the whole country.

And for another thing, it might not be fully possible for the rest of the country to emulate London's success. Insofar as better teachers would rather work in London than in Peterborough, poor whites in the latter are at a disadvantage. And as the JRF has complained, many schemes aiming to improve poor kids' achievements have been "a proliferation of 'hopeful' interventions with unknown effectiveness".

However, I don't mean this as a counsel of despair. I do so instead to repeat what I've said before - that if we want to improve the life-chances of the worst-off, the answer lies not just in investing in education, but in more income redistribution.